Skip to contentSkip to filters

We use cookies for analytics.

Aligning Evaluation Incentives To Reduce Hallucination

Recording

45 min

Abstract

Large language models are typically evaluated with accuracy-based metrics that reward confident guessing while giving little or no credit for acknowledging uncertainty. Because evaluations influence which models and methods are selected, these incentives can work against techniques that reduce hallucinations at the cost of answering fewer questions. The talk discusses evaluation approaches that better reward useful uncertainty and reflect the differing costs of mistakes, correct answers and abstentions, together with open questions in designing evaluations whose incentives better match real-world objectives.

Topics